Kubei : A Flexible Kubernetes Runtime Scanner

Kubei is a vulnerabilities scanning tool that allows users to get an accurate and immediate risk assessment of their kubernetes clusters. Kubei scans all images that are being used in a Kubernetes cluster, including images of application pods and system pods. It doesn’t scan the entire image registries and doesn’t require preliminary integration with CI/CD pipelines.

It is a configurable tool which allows users to define the scope of the scan (target namespaces), the speed, and the vulnerabilities level of interest.

It provides a graphical UI which allows the viewer to identify where and what should be replaced, in order to mitigate the discovered vulnerabilities.

Prerequisites

  • A Kubernetes cluster is ready, and kubeconfig ( ~/.kube/config) is properly configured for the target cluster.

Required Permissions

  • Read secrets in cluster scope. This is required for getting image pull secrets for scanning private image repositories.
  • List pods in cluster scope. This is required for calculating the target pods that need to be scanned.
  • Create jobs in cluster scope. This is required for creating the jobs that will scan the target pods in their namespaces.

Configurations

The file deploy/kubei.yaml is used to deploy and configure Kubei on your cluster.

  1. Set the scan scope. Set the IGNORE_NAMESPACES env variable to ignore specific namespaces. Set TARGET_NAMESPACE to scan a specific namespace, or leave empty to scan all namespaces.
  2. Set the scan speed. Expedite scanning by running parallel scanners. Set the MAX_PARALLELISM env variable for the maximum number of simultaneous scanners.
  3. Set severity level threshold. Vulnerabilities with severity level higher than or equal to SEVERITY_THRESHOLD threshold will be reported. Supported levels are Unknown, Negligible, Low, Medium, High, Critical, Defcon1. Default is Medium.
  4. Set the delete job policy. Set the DELETE_JOB_POLICY env variable to define whether or not to delete completed scanner jobs. Supported values are:
    • All – All jobs will be deleted.
    • Successful – Only successful jobs will be deleted (default).
    • Never – Jobs will never be deleted.

Usage

  • Run the following command to deploy Kubei on the cluster: kubectl apply -f https://raw.githubusercontent.com/Portshift/kubei/master/deploy/kubei.yaml
  • Run the following command to verify that Kubei is up and running: kubectl -n kubei get pod -lapp=kubei
  • Then, port forwarding into the Kubei webapp via the following command: kubectl -n kubei port-forward $(kubectl -n kubei get pods -lapp=kubei -o jsonpath='{.items[0].metadata.name}') 8080
  • In your browser, navigate to http://localhost:8080/view/ , and then click ‘GO’ to run a scan.
  • To check the state of Kubei, and the progress of ongoing scans, run the following command: kubectl -n kubei logs $(kubectl -n kubei get pods -lapp=kubei -o jsonpath='{.items[0].metadata.name}')
  • Refresh the page (http://localhost:8080/view/) to update the results.

Running Kubei with an external HTTP/HTTPS proxy

Uncomment and configure the proxy env variables for the Clair and Kubei deployments in deploy/kubei.yaml.

Amazon ECR support

Create an AWS IAM user with AmazonEC2ContainerRegistryFullAccess permissions.

Use the user credentials (AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY, AWS_DEFAULT_REGION) to create the following secret:

cat <<EOF | kubectl apply -f -
apiVersion: v1
kind: Secret
metadata:
  name: ecr-sa
  namespace: kubei
type: Opaque
data:
  AWS_ACCESS_KEY_ID: $(echo -n 'XXXX'| base64 -w0)
  AWS_SECRET_ACCESS_KEY: $(echo -n 'XXXX'| base64 -w0)
  AWS_DEFAULT_REGION: $(echo -n 'XXXX'| base64 -w0)
EOF

Note:

  • Secret name must be ecr-sa
  • Secret data keys must be set to AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY and AWS_DEFAULT_REGION

Google GCR Support

Create a Google service account with Artifact Registry Reader permissions.

Use the service account json file to create the following secret

kubectl -n kubei create secret generic –from-file=sa.json gcr-sa

Note:

  • Secret name must be gcr-sa
  • sa.json must be the name of the service account json file when generating the secret
  • Kubei is using application default credentials. These only work when running Kubei from GCP.

Limitations

R K

Recent Posts

Understanding the Model Context Protocol (MCP) and How It Works

Introduction to the Model Context Protocol (MCP) The Model Context Protocol (MCP) is an open…

5 days ago

The file Command – Quickly Identify File Contents in Linux

While file extensions in Linux are optional and often misleading, the file command helps decode what a…

5 days ago

How to Use the touch Command in Linux

The touch command is one of the quickest ways to create new empty files or update timestamps…

5 days ago

How to Search Files and Folders in Linux Using the find Command

Handling large numbers of files is routine for Linux users, and that’s where the find command shines.…

5 days ago

How to Move and Rename Files in Linux with the mv Command

Managing files and directories is foundational for Linux workflows, and the mv (“move”) command makes it easy…

5 days ago

How to Create Directories in Linux with the mkdir Command

Creating directories is one of the earliest skills you'll use on a Linux system. The mkdir (make…

5 days ago